Browser
How agents use pages
What an agent can do with a tab, and pages that need a password.
An agent drives its desk tabs through Sikemux’s browser_* tools. It can:
- read a page as numbered elements and text;
- click, type, press keys, scroll and drag;
- wait for a page to reach a state, such as text appearing or the network going quiet;
- take screenshots, draw on them and record the tab;
- read the network calls and console messages of the page.
You see each of these happen on the desk.
Pages that need a password
An agent never asks you to paste a password into its chat. When it needs one, it asks a secret question. You answer in a masked field, your answer goes straight to the agent as a file only your account can read, and the transcript shows only that it was answered.
Fields inside a frame from another site, such as a card field or a sign-in widget, can’t be read by the agent. It can still click into them and type at the caret.